arXiv：cs.LG（机器学习，全量分类）· Nafiseh Ghoroghchian, Haipeng Zhang, Shuyi Han, Alex Labach, George Stein·· 14 小时前AI 评分50
SimpleTimeBench 揭示时序基础模型在简单时间逻辑上的零样本盲区
Foundations without Fundamentals: Zero-Shot Blind Spots in Time Series FMs
AI 导读
arXiv 论文（arXiv:2610.02058）提出 SimpleTimeBench，用单调趋势、周期信号和先行指标协变量等单元测试诊断时序基础模型（TSFM）。
正文
Abstract:Despite the success of Time Series Foundation Models (TSFMs) on broad benchmarks, their ability to internalize basic temporal logic, especially in settings supported by exogenous covariates, remains under-examined. We introduce SimpleTimeBench, a diagnostic univariate and multivariate "unit test" suite for primitives such as monotonic trends, periodic signals and leading indicator covariates, scenarios where near-perfect forecasts should be trivial. Surprisingly, prominent multivariate TSFMs (Chronos-2, Moirai and Toto) frequently produce suboptimal zero-shot forecasts for these inputs. While fine-tuning Chronos-2 improves its behaviour on specific tasks, we show that this adaptation degrades performance on other fundamental patterns rather than enhancing its generalizable foundational capabilities. This reveals a gap between pre-training scale and basic temporal reasoning, suggesting that current TSFMs could potentially lack the inductive biases needed to capture simple predictable functions. We further demonstrate that these failures are not merely synthetic curiosities: they persist in real-world sensor forecasting, where TSFMs consistently underutilize leading indicators available in observed covariates. This inability to capture simple relationships limits the practical utility and reliability of current multivariate models.
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.02058 [cs.LG] |
| (or arXiv:2610.02058v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.02058 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Nafiseh Ghoroghchian [view email]
[v1]
Thu, 1 Oct 2026 17:02:16 UTC (2,068 KB)
来源：arXiv：cs.LG（机器学习，全量分类） · arxiv.org